[DATA] LogD data contains significant outliers

[hmacdope]

LogD data from ChEMBL has significant outliers, with values > 2000!!

So far I have clipped to [+- 10 units] downstream before using in workflows but this should probably be folded into the curation itself.